def getMelbourneSuburbs(file="Australian_Post_Codes_Lat_Lon.csv", maxDistance = 30):
'''Reads a file containing postcodes and locations (lat, long) and returns a DataFrame containing suburns within 30 km of Melbourne CBD.'''
postcodes = pd.read_csv(file)
postcodes = postcodes.query('state == "VIC"' )
postcodes['inROI'] = postcodes[ ['lat', 'lon'] ].apply(lambda x: isPointInROI(x[0], x[1], maxDistance = maxDistance) , axis = 1)
postcodes = postcodes[(postcodes.inROI) & (postcodes['postcode'] < 3400) & (postcodes['postcode'] != 3409)]
return(postcodes)
def isPointInROI(lat, long, maxDistance = 30):
'''Checks wheather a point on map falls within a square region of interest.
length of the ROI is maxDistance x 2, with ref coord of melbourne as center (0,0).
1 degree on lat and long is roughly 110 km.
'''
return((abs(lat - Melbourne_Center_Coord['lat']) < (maxDistance/110)) & (abs(long - Melbourne_Center_Coord['long']) < (maxDistance/110)))
def parseJson(json_raw, plotGraph = True, verbose = True):
import json
import re
beds = int(re.findall('\d{1}bd_', json_raw)[0].replace("bd_",""))
# json_raw = 'data/Prahran_VIC_3181_2bd_house.json'
f = open (json_raw, "r")
text = f.read()
f.close()
if (len(re.findall('400 Bad Request', text)) == 0 and len(re.findall('Internal Server Error', text)) == 0 ):
json_list = json.loads(text)
rawList = json_list['series']['seriesInfo']
list_df = []
for i in rawList:
temp_dict = {}
temp_dict['month'] = i['month']
temp_dict['year'] = i['year']
temp_dict['year_month'] = str(i['year']) + "_" + str(i['month'])
for key, value in i['values'].items():
temp_dict[key] = value
list_df.append(temp_dict)
df = pd.DataFrame(list_df)
df['suburb'] = json_list['header']['suburb']
df['propertyCategory'] = json_list['header']['propertyCategory']
df['bedrooms'] = beds
if plotGraph:
plt.plot(df['year'],df['medianSoldPrice'], 'ob-')
plt.title(str(beds) + " bedroom " + json_list['header']['propertyCategory'] + ", " + json_list['header']['suburb'] + ' (Annual data)' )
plt.show()
return(df)
else:
if (verbose): print("Insufficient data for:",json_raw)
return([""])
def calculateGrowth_wide(df, factor = 'medianSoldPrice'):
assert factor in df.columns, "Column " + factor + " not found in df"
df = df[~df[factor].isnull()] # remove Nan and see remaining data
if(df.shape[0] > 2 ): # num of non nan rows
Y = np.log10(np.array(df[factor]).reshape(-1,1))
# X = np.array(df.index).reshape(-1,1) # same order as df.year_month
X = np.array(df.year - 2010).reshape(-1,1)
reg = LinearRegression().fit(X = X, y = Y)
stats_dict = {'suburb' : list(df.suburb.unique()),
'propertyCategory' : list(df.propertyCategory.unique()),
'bedrooms' : list(np.int_(df.bedrooms.unique())),
# 'Type' : factor,
factor + '_R-squared': round(reg.score(X, Y),2),
factor + '_log10_growth' : list(np.round(reg.coef_[0],4)),
factor + '_log10_intercept' : list(np.round(reg.intercept_,2)),
factor + '_linear_growth' : [round(10**c,4) for c in list(reg.coef_[0])],
factor + '_percent_growth' : [100*(round(10**c,4) - 1) for c in list(reg.coef_[0])],
# factor + '_linear_intercept' : [int(10**c) for c in list(reg.intercept_)],
factor + '_N' : df.shape[0],
factor + '_starting_price' : list(df[df.year == df.year.min()].loc[:,factor]),
factor + '_final_price' : list(df[df.year == df.year.max()].loc[:,factor]) }
else:
stats_dict = {'suburb' : list(df.suburb.unique()),
'propertyCategory' : list(df.propertyCategory.unique()),
'bedrooms' : list(np.int_(df.bedrooms.unique())),
# 'Type' : factor,
factor + '_R-squared': None,
factor + '_log10_growth' : None,
factor + '_log10_intercept' : None,
factor + '_linear_growth' : None,
factor + '_percent_growth' : None,
# factor + '_linear_intercept' : None,
factor + '_N' : df.shape[0],
factor + '_starting_price' : None,
factor + '_final_price' : None}
return(pd.DataFrame(stats_dict))
## calculate distance from Melbourne
def distance_cbd(lat, long, region = 'Melbourne'):
major_city_coord = {'Melbourne' : {'lat': -37.814218,'long': 144.963526},
'Sydney' : {'lat': -33.87003635458607, 'long': 151.20745239732523},
'Parramatta' :{'lat': -33.815101579800746, 'long': 150.99980625492702}
}
assert region in major_city_coord, "Location must be one of :" + major_city_coord.keys()
'''Checks wheather a point on map falls within a square region of interest.
length of the ROI is maxDistance x 2, with ref coord of melbourne as center (0,0).
1 degree on lat and long is roughly 110 km.
'''
return( np.sqrt(np.power(lat - major_city_coord[region]['lat'], 2) + np.power(long - major_city_coord[region]['long'], 2)))
def drawPropertyMap(df, propertyCategory = "House", bedrooms = 2, maxFinalPrice = 10**9, showStandardised = True):
q = 'propertyCategory == "' + propertyCategory +'" & bedrooms == ' + str(bedrooms) + ' & medianSoldPrice_final_price <= ' + str(maxFinalPrice)
print(q)
# df.query(q).shape
geo_df = geopandas.GeoDataFrame(df.query(q))
print("Results: " + str(geo_df.shape[0]))
plotTitle = str(bedrooms) + ' bedroom ' + propertyCategory + ' under $' + '{:,.0f}'.format(maxFinalPrice)
#! populate cursor hover text
my_customData2 = []
for i in range(geo_df.shape[0]):
temp_slice = geo_df.iloc[i, ::]
my_text = temp_slice['suburb'].upper() + \
', Property appreciation rate: '+ str(temp_slice['medianSoldPrice_percent_growth']) + \
', N =' + str(temp_slice['medianSoldPrice_N']) + \
', R^2 = ' + str(temp_slice['medianSoldPrice_R-squared']) + \
', Current median price:$ ' + str(temp_slice['medianSoldPrice_final_price']) + \
', Current median Rent: $' + str(temp_slice['medianRentListingPrice_final_price']) + \
', Rental growth: ' + str(temp_slice['medianRentListingPrice_percent_growth']) +'% pa'
my_customData2.append(my_text)
## plotting
lga_json = geo_df.__geo_interface__
if showStandardised:
geo_df['medianSoldPrice_percent_growth_standardised'] = scipy.stats.zscore(list(geo_df['medianSoldPrice_percent_growth']))
fillValues = geo_df['medianSoldPrice_percent_growth_standardised']
zmin = geo_df['medianSoldPrice_percent_growth_standardised'].min()
zmax = geo_df['medianSoldPrice_percent_growth_standardised'].max()
my_hoverTemplate = "<b>Std Growth: %{text} </b><br>" + '%{customdata}' + "<extra></extra>"
my_colour_palette = 'PRGn'
else:
fillValues = geo_df['medianSoldPrice_percent_growth']
zmin = geo_df['medianSoldPrice_percent_growth'].min()
zmax = geo_df['medianSoldPrice_percent_growth'].max()
my_hoverTemplate = "<b>Growth: %{text} </b><br>" + '%{customdata}' + "<extra></extra>"
my_colour_palette = 'Blues'
# print(lga_json)
# print(geo_df.suburb)
# Set the data for the map
data = go.Choroplethmapbox(
geojson = lga_json, #this is your GeoJSON
locations = geo_df.index, #the index of this dataframe should align with the 'id' element in your geojson
z = fillValues, #sets the color value
customdata = my_customData2,
text = fillValues, #sets text for each shape
colorbar = dict(thickness=20, ticklen=3,outlinewidth=0), #adjusts the format of the colorbar
marker_line_width=1, marker_opacity=0.7,
colorscale= my_colour_palette , # Viridis_r adjust format of the plot
zmin = zmin, zmax = zmax, #sets min and max of the colorbar
hovertemplate = my_hoverTemplate) # sets the format of the text shown when you hover over each shape
# Set the layout for the map
layout = go.Layout(
title = {'text': plotTitle,
'font': {'size':24}}, #format the plot title
mapbox1 = dict(
domain = {'x': [0, 1],'y': [0, 1]},
center = dict(lat=-37.814218 , lon=144.963526),
accesstoken = MAPBOX_ACCESSTOKEN,
zoom = 8.5),
autosize=True,
height=600,
margin=dict(l=0, r=0, t=40, b=0))
# Generate the map
fig=go.Figure(data=data, layout=layout)
fig.show()
